Multiple kinesin-2 subtypes traverse entire OSN axonemes

Our data highlight structural and membrane compositional aspects distinguishing specific subdomains of OSN cilia, but the mechanisms by which the proximal and distal segments are established remain unclear. In C. elegans amphid/phasmid channel sensory cilia, two kinesin motors function in coordination to build the proximal/middle and distal ciliary axoneme segments37. To examine whether multiple kinesin motors are used in OSN cilia, we expressed FP-tagged homodimeric Kif17 and a component of the heterotrimeric kinesin-2 (Kap3a). Both Kap3a and Kif17 localized in puncta along the full length of OSN axonemes and showed accumulations at ciliary tips (Fig. 3a–d). Thus, heterotrimeric kinesin-2 is not restricted to the proximal segment of OSN cilia as seen in nematode amphid/phasmid channel cilia. FP-tagged Dync2li1, a component of the ciliary retrograde dynein motor complex, showed similar localization (Fig. 3e,f). To assess ciliary movement of the motor complex proteins, we employed high contrast, rapid acquisition, total internal reflection fluorescence microscopy (TIRFm) on OSNs expressing Kap3a:GFP, Kif17:mCitrine or Dync2li1:GFP. Strikingly, each protein showed robust, processive particle movement in both the anterograde and retrograde directions along the entire OSN cilia (Fig. 3g–k and Supplementary Movie 1). To our knowledge, this is the first capture of IFT in any native tissue of a vertebrate and/or mammalian system.

Open in a separate window

Figure 3

Homodimeric and heterotrimeric kinesin-2 motors cooperatively traffic along the full length of OSN cilia.

(a–c) Representative live en face confocal images of native OE ectopically expressing components of the IFT motor complexes. (a) Homodimeric kinesin-2 protein Kif17:mCitrine is present in cilia and at ciliary distal tips. (b) Accumulations of Kif17:mCitrine at distal tips (arrowheads) and in puncta along axonemes (arrow). (c) Heterotrimeric kinesin-2 component Kap3a:GFP is abundant along cilia axonemes. (d) Accumulations of Kap3a:GFP at ciliary distal tips (arrowheads) and distribution along axonemes (arrows). (e) Cytoplasmic dynein component Dync2li1:GFP is detectable along cilia axonemes. (f) Dync2li1:GFP is highly abundant along OSN cilia, detectable as discrete puncta (arrows), and is present at ciliary distal tips (arrowheads). Bracket indicates signal in a cilium from an OSN expressing only AV-Dync2li1:GFP. (g) TIRFm time-lapse capture of Kap3a:GFP particle movement in an OSN cilium. Exposures from different time points were overlayed onto an average intensity projection of the entire series to resolve individual particles in relation to the axoneme (ciliary tip is to the right, top). (Arrows) Anterograde particles. (Arrowheads) Retrograde particles. (h–k) Line-scan kymograms generated from single cilia of OSNs ectopically expressing the indicated motor component. (h) Orientation of particle movement in relation to the ciliary tip. Axes represent distance (μm) over time (s). (i) Kif17:mCitrine-, (j) Kap3a:GFP- and (k) Dync2li1:GFP-labelled particles moving in anterograde and retrograde directions. (l) Disruption of Kif17 is not detrimental to the maintenance of OSN axonemes. Representative live en face confocal image of an OSN co-expressing AV-Kif17DN:GFP and AV-Arl13b:mCherry. (m) The presence of Kif17DN:GFP at distal tips (arrowheads) and in puncta along axonemes (arrows). (n,o) Kif17DN is transported to distal tips in IFT particles containing Kap3a. (n) Representative live en face confocal image of an OSN co-expressing AV-Kif17DN:mCherry and AV-Kap3a:GFP shows their co-localization in cilia and at ciliary distal tips (arrowheads). (o) Kymogram generated from a cilium of an OSN expressing (top) AV-Kif17DN:GFP and (middle) AV-Kap3a:mCherry. Scale bars, 10μm (a–e,l,n); 2.5μm (zoomed panels); 2.5μm (g); 10μm × 30s (h–k,o).

Our time-lapse data indicates that both heterotrimeric kinesin-2 and Kif17 participate in IFT along the entire OSN axonemes. We therefore reasoned that the presence of OSN cilia distal segments may not depend solely on the function of Kif17 as is seen with OSM-3 in C. elegans amphid/phasmid channel cilia37. To test this hypothesis, we used a truncated dominant-negative version of Kif17 (Kif17DN) lacking the motor domain. Expression of Kif17DN was not detrimental to OSN cilia formation (Fig. 3l,m), and Kif17DN showed mobility along the entire axonemes via association with IFT particles containing Kap3a (Fig. 3n,o and Supplementary Movie 2). These data indicate that distal segment formation and/or maintenance in OSN cilia is not dependent on Kif17 alone, and that heterotrimeric kinesin-2 is capable of IFT trafficking along the entire length of OSN axonemes when Kif17 is disrupted. Thus, we conclude that OSN cilia use a unique and heretofore undescribed programme featuring at least two types of kinesin motors for trafficking along distal singlet microtubules.

Stochastic variation in IFT velocities among OSNs

The basic mechanistic principles by which IFT trains enable cargo transport within cilia are well established from work in homogeneous age-synchronous cell populations or groups of cells derived from an invariant lineage, but we have no understanding about IFT dynamics in the complex setting of a native mammalian tissue or in multiciliated cells. As individual OSNs differ from each other on many levels including odourant receptor expression, basal stimulation/activation state, age and spatial distribution, the OE is well suited for analysis of IFT dynamics in a heterogeneous context. We measured IFT velocities using FP-tagged components of the IFT-A and IFT-B subcomplexes, which undergo robust bidirectional motility along the full length of OSN cilia (Fig. 4a–h and Supplementary Movie 3). To differentiate anterograde versus retrograde trajectories, only IFT particles travelling in clear view of the ciliary tip were included in our analysis and are representative of IFT velocities specifically on the OSN distal segment. Velocities of both anterograde and retrograde IFT particles ranged from nearly stationary to approaching 0.6μms⁻¹ (Fig. 4i), with anterograde and retrograde means of 0.22±0.08 and −0.14±0.08μms⁻¹, respectively (n=2,316 IFT122:GFP-labelled particles, 1,941 IFT88:GFP-labelled particles, 300 cilia, 55 OSNs and 9 mice). In addition, the magnitude of anterograde and retrograde mean particle velocities within individual OSN cilia strongly correlate (Fig. 4j, Pearson’s coefficient R²=0.6064). Notably, we saw stochastic variation in anterograde and retrograde velocities among neurons and the velocities measured in multiple cilia within individual cells were more similar to each other than to cilia on other cells (Fig. 4k; permutation test with 20,000 iterations, P<0.0001 for both anterograde and retrograde velocities, see Methods) (n=1,941 IFT88:GFP-labelled particles, 160 cilia, 25 OSNs and 5 mice). Thus, differential regulation of IFT velocity occurs across the OSN population and the velocities of IFT particles in the many cilia projecting from a single OSN are controlled together within each cell.

Open in a separate window

Figure 4

Stochastic variation in IFT velocities among OSNs.

(a,c,e) Representative live en face confocal images of native OE ectopically expressing components of the IFT particle subcomplexes. Localization of (A) IFT-A subcomplex protein IFT122:GFP and (c) IFT-B subcomplex protein IFT88:GFP in the OSN knob and in puncta along the length of cilia as revealed by Arl13b:mCherry co-expression. Zoomed colour-shifted images show accumulation of (b) IFT122:GFP and (d) IFT88:GFP at the distal tips of cilia (arrow) and in puncta along axonemes (arrowheads). (e,g) Representative live en face TIRFm images captured from OSNs expressing IFT122:GFP or IFT88:GFP. Two-second exposures are shown. Insets show GFP puncta on OSN ciliary axonemes (arrowheads) and accumulations at ciliary distal tips (arrows). (f,h) Representative line-scan kymograms generated from single cilia of OSNs ectopically expressing the indicated IFT protein. (i) Histogram distribution of IFT122:GFP and IFT88:GFP particle velocities. IFT122:GFP, n=1,315 anterograde particles and 1,001 retrograde particles, 4 mice; IFT88:GFP, n=1,060 anterograde particles and 881 retrograde particles, 5 mice. (j) Plot of individual cilium mean anterograde versus retrograde velocities, showing a positive association between the magnitude of anterograde and retrograde velocities in a given cilium (Pearson’s coefficient R²=0.6064). Linear regression is shown. (k) Plot of individual cilium mean anterograde versus retrograde velocities, clustered by OSN (each group of points encapsulated by a coloured oval represents the cilia from a single cell). IFT88:GFP data set is shown. Scale bars, 10μm (a,c); 2.5μm (b,d); 10μm, inset, 2.5μm (e,g); 5μm × 10s (f,h).

Although the majority of IFT particles in OSN cilia moved with constant velocity, in some instances we observed non-processive particle trajectories (Supplementary Fig. 3). These included single-particle velocity changes (stalling/restarting and mid-cilium direction reversal), which may represent spatio-temporal events where IFT particles perform maintenance on the axoneme through delivery and/or pickup of cargo38. We also observed unidirectional particle interactions (fusions and fissions) and rare anterograde and retrograde particle collision/fusion events. These observations imply that individual microtubules may be simultaneously used by both unidirectional and opposing particles, and that particle remodelling is not strictly limited to events at the ciliary base and microtubule tips39. Finally, increased particle size was occasionally observed, which may reflect a form of IFT avalanching at the ciliary base or tip as described in C. reinhardtii40 and/or snowballing from multiple fusion events along the axoneme.

The BBSome is a constituent of the mammalian OSN IFT complex

Proteins homologous to those associated with human BBS participate in IFT in invertebrates41,42,43. Although various lines of evidence have implicated mammalian BBS proteins as being resident within pericentriolar satellites44, basal bodies45 and cilia46, no report has demonstrated that mammalian BBS proteins traffic with IFT. We examined constituents of the core BBSome46 via expression of FP-tagged BBS1, BBS2 and BBS4 in OSNs. Confocal microscopy revealed strong enrichment of BBSome proteins within OSN knobs (Fig. 5a–f), consistent with basal body localization. In addition, BBS2:GFP was readily detectable in puncta along OSN axonemes (Fig. 5c inset), whereas BBS1:GFP was discretely present in the proximal-most regions of OSN axonemes (Fig. 5b) and BBS4:GFP was occasionally detected at ciliary tips (Fig. 5g). Using TIRFm, we observed that each of the three BBSome fusion proteins undergo bidirectional particle movement along the length of OSN cilia (Fig. 5j–l and Supplementary Movie 4) and do so with velocities similar to FP-tagged IFT proteins (Fig. 5m and Table 1; n=2,881 BBS4:GFP-labelled particles, 143 cilia, 48 OSNs and 7 mice). We next co-expressed BBS4:mCherry and IFT88:GFP, and found that BBS4:mCherry was present in all observed IFT88:GFP-labelled particles (Fig. 5n and Supplementary Movie 5), verifying that the BBSome is in fact a constituent of the IFT particle in OSN cilia.

Open in a separate window

Figure 5

The BBSome is a constituent of OSN IFT particles.

(a–f) Representative live en face confocal images of native OE ectopically expressing components of the BBSome. (a) Image of OE from animal transduced with AV-BBS1:GFP and AV-Arl13b:mCherry. BBS1:GFP concentrates strongly within the OSN knob and is distributed uniformly along the proximal-most segments (arrow) of OSN cilia as shown in the zoomed image (a). (c,d) Image of OE from animal transduced with AV-BBS2:GFP and AV-Nphp4:mCherry. (c) BBS2:GFP concentrates strongly within the OSN knob and is visible along ciliary axonemes in puncta (inset, arrowheads). (d) BBS2:GFP is enriched in basal bodies in the OSN knob. Merged panel shows BBS2:GFP basal body localization in close association with transition zones (arrows) as revealed by Nphp4:mCherry labelling. (e) Image of OE from animal transduced with AV-BBS4:GFP and AV-Arl13b:mCherry. BBS4:GFP concentrates strongly within the OSN knob. (f) Zoomed panel showing a transduced OSN knob where BBS4:GFP appears to localize at basal bodies. (g) Merged and zoomed panel shows accumulation of BBS4:GFP at the tip of an OSN cilium as revealed by Arl13b:mCherry labelling. (h–n) TIRFm uncovers BBSome protein association with IFT particles in OSN cilia. (h,i) Collapsed maximum intensity projection of TIRFm time-series images captured from OSNs ectopically expressing (h) AV-BBS1:GFP or (i) AV-BBS4:GFP. The collapsed time series show that both BBS1:GFP and BBS4:GFP are present in OSN cilia. (j–l) BBSome proteins show bidirectional particle movement in OSN cilia. Line-scan kymograms of cilia on OSNs ectopically expressing (j) AV-BBS1:GFP, (k) AV-BBS2:GFP or (l) AV-BBS4:GFP were generated from TIRF time-series acquisitions from native OE. (m) Histogram distribution of BBS4:GFP particle velocities. n=1,735 anterograde particles and 1,146 retrograde particles. (n) BBS4:mCherry associates with IFT particles. Representative kymogram was generated from a cilium of an OSN co-expressing AV-BBS4:mCherry and AV-IFT88:GFP. Scale bars, 10μm (a,c,e); 2.5μm (inset of c); 2.5μm (b,d,f,g); 10μm (h,i); 5μm × 10s (j–l,n).

Table 1

Comparison of IFT velocities across species and cell types.

Species/cell type	Marker	Length (μm)	Anterograde (μms−1)	Retrograde (μms−1)	Citation
C. reinhardtii	DIC	12	1.8	3.1	Iomini et al.67
C. reinhardtii	IFT20:GFP	12	2.1	2.5	Lechtreck et al.43
C. reinhardtii	IFT27:GFP	2–4	~1.6	nd	Engel et al.68
C. reinhardtii	IFT27:GFP	6–8	~2.0	nd	Engel et al.68
C. reinhardtii	IFT27:GFP	10+	~2.3	nd	Engel et al.68
Trypanosoma brucei (27°C)	GFP:IFT52, GFP:DHC2.1	22.3	2.4 (1.5)*	5.6	Buisson et al.39
T. brucei (35°C)	GFP:IFT52, GFP:DHC2.1	22.3	3.2 (2.2)*	7.4	Buisson et al.39
C. elegans/Amphid-middle	OSM-6:GFP	4	0.68	1.2	Snow et al.37
C. elegans/Amphid-distal	OSM-6:GFP	2.5	1.3	1.1	Snow et al.37
Mus musculus/LLC-PK1	IFT20:GFP	~6	~0.6	~0.7	Follit et al.14
M. musculus/IMCD3	IFT88:YFP	6	0.38	0.56	Tran et al.13
M. musculus/MEF	IFT88:GFP	~3	1.09	0.72	He et al.18
M. musculus/OSN-distal	IFT122:GFP	2.5–100+	0.22	0.14	Current work
M. musculus/OSN-distal	IFT88:GFP	2.5–100+	0.23	0.14	Current work
M. musculus/OSN-distal	Kif17:mCitrine	2.5–100+	0.23	nd	Current work
M. musculus/OSN-distal	Kif17DN:GFP	2.5–100+	0.23	nd	Current work
M. musculus/OSN-distal	BBS4:GFP	2.5–100+	0.24	0.17	Current work

Open in a separate window

Nd, not determined.

^*Subset of slow particle velocities.

Assembly of the BBSome depends up the activity of a BBS–chaperonin complex containing BBS6, BBS10 and BBS12 (ref. 47). As BBS10 and BBS12 are present at or near basal bodies48 where the BBSome is most abundant in primary ciliated cells, we hypothesized that they may also associate with the BBSome during IFT in OSN cilia. Confocal analysis revealed BBS10:GFP localization almost exclusively in OSN knobs (Supplementary Fig. 4). Within the OSN knob, however, BBS10:GFP did not associate with all basal bodies as is typical for BBS proteins; instead, BBS10 accumulated as a single focus (Supplementary Fig. 4b–e). Remarkably, the BBS10:GFP focus was highly dynamic, displaying constant mobility (Supplementary Fig. 4f). It is enticing to speculate that BBS10 and other components of the BBS–chaperonin complex are resident within a single pericentriolar endosome that actively shuttles throughout the OSN knob to provide support to each olfactory cilium. Using TIRFm on OSNs co-expressing BBS10:GFP and BBS4:mCherry, we observed only rare, transient association of BBS10 with IFT particles (Supplementary Fig. 4g). Thus, we conclude that the major function of the BBS–chaperonin complex in BBSome assembly is performed outside of OSN cilia before incorporation of the BBSome into IFT particles.

Peripheral membrane Arl proteins are OSN IFT cargoes

We next examined peripheral ciliary membrane Arl proteins. Arl6 (hereafter referred to as BBS3) is required for BBSome recruitment onto membranes in vitro49 and for localization of the BBSome within mammalian cilia on cultured cells50. In C. elegans, BBS3 and Arl13b homologues move bidirectionally within cilia, suggesting they associate with IFT41,51. Similar to Arl13b:mCherry, we observed BBS3:GFP throughout the full length of OSN axonemes (Fig. 6a,c). Remarkably, both BBS3:GFP and Arl13b:mCherry showed discrete bidirectional ciliary transport when imaged by TIRFm (Fig. 6d–g and Supplementary Movie 6). To assess the stoichiometry of BBS3 and Arl13b association with IFT particles, we co-expressed each protein with FP-tagged BBS4 and performed dual colour time-lapse TIRFm. Although BBS3:GFP appeared to co-localize with all BBS4:mCherry-labelled particles (Fig. 6f), Arl13b:mCherry was clearly absent from most (69.2%) BBS4:GFP particles (n=340) (Fig. 6g and Supplementary Movie 7), indicating that Arl13b is an intermittent IFT cargo and not a core constituent of the IFT machinery.

Open in a separate window

Figure 6

Peripheral ciliary membrane-associated Arl proteins participate in IFT.

(a–c) Representative live en face confocal images of native OE ectopically co-expressing AV-BBS3:GFP and AV-Arl13b:mCherry. (a) BBS3:GFP is present along OSN ciliary axonemes. Inset shows zoomed image of two axonemes decorated with BBS3:GFP. (b) Zoomed image of the OSN knob showing BBS3:GFP accumulation. (c) Zoomed and colour-shifted image of an OSN ciliary distal tip showing co-localization of BBS3:GFP and Arl13b:mCherry. (d) Line-scan kymogram generated from a TIRFm time series of a BBS3:GFP AV-transduced OSN showing bidirectional movement of BBS3:GFP in an OSN cilium. (e) Line-scan kymogram generated from a TIRFm time series of an Arl13b:mCherry AV-transduced OSN showing particle movement of Arl13b:mCherry in an OSN cilium. (f) Line-scan kymogram of an OSN cilium co-expressing AV-BBS3:GFP and AV-BBS4:mCherry. (g) Line-scan kymogram of an OSN cilium co-expressing AV-Arl13b:mCherry and AV-BBS4:GFP. White arrows highlight a particle possessing both Arl13b:mCherry and BBS4:GFP. Black arrow indicates a group of particles that are labelled by BBS4:GFP but not Arl13b:mCherry. Scale bars, 10μm (a); 2.5μm (a (inset), b,c); 5μm × 10s (d–g).

Scanning electron microscopy

After deep anaesthesia and transcardial perfusion with 2% glutaraldehyde, 0.15M cacodylate in water, olfactory tissue was processed using the OTOTO protocol. Analysis was performed on an Amray (Drogheda, Ireland) 1910FE field emission scanning electron microscope at 5kV and recorded digitally with Semicaps software.

Constructs

Plasmids containing complementary DNA inserts used in this study were generously provided as follows: Nphp4:mCherry and IFT88 (B.K.Y., UAB); GFP:Efhc1 (K.Y., RIKEN); Arl13b (T.C., Emory); IFT122:GFP (J.E., UGA); Kap3a (T.S., Johns Hopkins); Olfr78 (J.P., Johns Hopkins); ACIII (R.R., Johns Hopkins); BBS1 (V.C.S., Iowa); and BBS2, BBS3, BBS4 and BBS10 (K.M., Ohio State). Dync2li1 was amplified from cDNA generated from mouse whole eye RNA. Lipid-anchored GFP constructs were generated by annealing oligonucleotides encoding the 13 amino-terminal residues from Lyn kinase (MyrPalm), the 20 N-terminal residues from GAP43 (PalmPalm) and the 9 carboxy-terminal residues from the guanine triphosphatase Rho (GerGer) into the KpnI and AgeI sites of pEGFP-N1 (Clontech). All cloning primer sequences are available on request.

Analysis of ectopic expression in mouse OSNs

For expression in native tissue, recombinant GFP-, mCherry- or mCitrine-flanked cDNAs were inserted into the adenoviral vector pAD/V5/−dest and virus was propagated using the ViraPower protocol (Invitrogen). AV particles were isolated with the Virapur Adenovirus mini purification Virakit (Virapur, San Diego, CA) and dialysed in 2.5% glycerol, 25mM NaCl and 20mM Tris-HCl, pH 8.0. Twenty-microlitre volumes of AV-containing constructs of interest were intranasally administered to the OE of pups at P7, P8 and P9. Transduced animals were killed for analysis on or after day P19.

For immunostaining, mice were deeply anaesthetized and then cardiac perfused with 4% paraformaldehyde. The lower jaw, teeth and hair were removed from snouts before postfixation in 4% paraformaldehyde for 2h at 4°C. Snouts were then decalcified in 0.5M EDTA/ × PBS overnight, cryoprotected in 10%, 20% and 30% sucrose for 1h, 1h and overnight, respectively, at 4°C, and embedded in optimal cutting temperature compound (Tissue-Tek). Coronal cryosections were sliced at a thickness of 10–12μm and mounted onto Superfrost Plus slides (Fisher Scientific). For immunostaining, cryosections were permeabilized and blocked with 0.3% Triton X-100 and 2% goat serum in PBS for 30min. Samples were incubated with acetylated α-tubulin antibody (clone 6–11 B-1, Sigma T6793) diluted 1:1,000 in 2% goat serum in PBS for 1h at room temperature and washed three times with PBS. Cryosections were incubated in fluorescent-conjugated secondary antibodies for 1h, washed three times with PBS and mounted using Prolong Gold (Invitrogen). Fixed tissue imaging was performed on an Olympus Fluoview 500 confocal microscope equipped with a 405-nm laser diode with a 430- to 460-nm band-pass filter, a 488-nm laser with a 505- to 525-nm band-pass filter, a 543-nm laser with a 560-nm long-pass filter and a 633-nm laser with a 660-nm long-pass filter; original magnification × 60, 1.40 numerical aperture (NA) or × 100,1.35 NA oil objectives were used.

For confocal en face imaging of transduced OSNs, animals were anaesthetized with CO₂, rapidly decapitated, split along the cranial midline and the OE was dissected out in PBS. Turbinates were placed in a tissue chamber and imaged on a Nikon TiE-PFS-A1R confocal microscope equipped with a 488-nm laser diode with a 510- to 560-nm band-pass filter and 561 laser with a 575- to 625-nm band-pass filter. A CFI Apo Lamda S × 60, 1.4 NA objective was used.

For en face TIRFm imaging of transduced OSNs, animals were anaesthetized with CO₂, rapidly decapitated, split along the cranial midline and further prepared in artificial cerebrospinal fluid (124mM NaCl, 3mM KCl, 1mM MgCl₂, 2mM CaCl₂, 1.25mM NaH₂PO₄, 26mM NaHCO₃, 25mM glucose) bubbled with 5% CO₂/95% O₂ for 10min before use. The rostral- and caudal-most portions of one hemisphere were removed, leaving the OE and olfactory bulb intact in the skull. The tissue was placed face down in the imaging chamber, bathed in 35°C artificial cerebrospinal fluid and held in place with mesh netting. TIRFm time series were captured at 200ms exposure for 2–3min on a Nikon Eclipse Ti-E/B inverted microscope equipped with a × 100 CFI APO TIRF 1.49 NA, × 1.5 tube lens, ZT488/561rpc dichroic, ZET488/561 × excitation filter, ZET488/561m-TRF emission filter (Chroma Technology) and an electron-multiplying charge-coupled device camera (iXon X3 DU897, Andor Technology). The 488-nm line of a fibre-coupled diode laser at incident power of 2mw was used to illuminate a circular region of ~60μm in diameter for capturing video sequences at 5–10Hz. ImageJ was used to generate line-scan kymograms for measuring particle velocities from imported time series. Velocity was calculated with the following formula: tan(α*π/180)*b/c where α is angle, b is calibration in μm per pixel and c is exposure time per frame.

We thank L. Beyer and members of the University of Michigan Microscopy and Image Analysis Laboratory for assistance with scanning electron microscopy. We thank H.L. Kee, J. Pieczynski and additional members of the Martens and Verhey labs for valuable discussions and technical assistance. This work was supported by NIH grants R01DC009606 (to J.R.M.), RO1070862 (to K.V.), T32DC00011 (to C.L.W.) and K99DC013555 (to J.C.M.).